MultiNav Pro+ BLE - does RSSI scale predictably across distance?

Michael Johnson · 2026-06-23T20:53:39.000Z

Working on a ranging/proximity system using the MultiNav Pro+ BLE modules and I'm trying to figure out if I can use RSSI as a rough distance estimator between two nodes. I know RSSI-based ranging is never super accurate but for my use case ±5-10 meters is totally fine - I just need a ballpark. The question is whether the RF front-end on this thing behaves predictably enough that a standard log-distance path loss model would actually fit the RSSI curve, or if the front-end does something funky with gain control that messes with the raw RSSI values at different distances.

Replies

Patricia Davis•June 23, 2026

RSSI on the MultiNav Pro+ is reasonably well-behaved for log-distance modeling in my experience - I've gotten decent fits with a standard path loss exponent around 2.5-3.0 indoors, and for ±5-10m accuracy you should be fine as long as you calibrate your reference RSSI at 1m first. The one thing I'd watch out for is the onboard antenna's gain pattern isn't super uniform so orientation of the nodes matters more than you'd expect, especially if they're not both upright.

Adam•June 23, 2026

@Michael Johnson yeah this is a reasonable thing to try, and ±5-10m is a realistic target for RSSI-based ranging so you're not setting yourself up for disappointment.

The honest answer on the front-end behavior: the amplifier stage on the MultiNav Pro+ is what gives it the extended range, and external LNAs can introduce some non-linearity at very close distances (roughly sub-5m) where the signal is strong enough that the amp starts compressing. That can flatten the RSSI curve at the near end, which throws off a naive log-distance fit if you're calibrating across a wide range. The mid-to-far range (say 10m out to a few hundred meters) tends to behave much more predictably and fits the standard model reasonably well.

Practical suggestion: do your calibration measurements starting around 3-5m rather than 1m, collect a decent sample at each distance increment (RSSI is noisy, you want to average at least 10-20 readings per point), then fit your path loss exponent from that data rather than assuming n=2. Indoors you'll probably land somewhere between 2.5-3.5 depending on your environment, outdoors closer to 2. Once you have your empirical curve the ±5-10m target should be achievable in reasonably open conditions.

If you end up sharing your calibration data in the Dev Hub I'd be curious to see how the curve looks — would help others doing similar proximity work too.

Richard Rodriguez•June 23, 2026

Agreed with Patricia on the orientation thing - that bit me hard on a first prototype where one node was mounted horizontally and my RSSI curve was all over the place until I figured out what was happening. Also worth averaging a few samples before feeding into your path loss calc, the MultiNav's BLE frontend can be a bit jittery on a single reading especially if theres any 2.4GHz congestion nearby.

Thomas Jones•June 23, 2026

One more thing worth adding - if you're in an environment with any reflective surfaces (metal shelving, concrete walls, etc.) you might see your path loss exponent creep up toward 3.5 or even higher, so I'd actually do your 1m calibration in the same physical environment you're deploying in rather than open-air. Took me an embarrassing number of iterations to figure out why my model kept drifting between test bench and real install.

Sarah Taylor•June 23, 2026

All solid advice above - one small thing I'd add is to throw a median filter on your RSSI samples rather than a straight average, it handles the occasional wild outlier from BLE interference way better and made my distance estimates noticeably more stable without needing a ton of extra samples.

Joseph Brown•June 23, 2026

All great points in this thread - one thing I'd also throw in is that if you're polling RSSI at a high rate you might actually be hurting yourself, slowing down to maybe 5-10 samples per second and letting the median filter do its thing tends to give cleaner results than hammering it and averaging a ton of noisy readings. Also make sure your TX power is locked in and not doing anything adaptive, had a fun afternoon once where my path loss model made zero sense and it turned out the module was quietly adjusting TX power on me.

Susan Gonzalez•June 23, 2026

Late to this thread but wanted to confirm the log-distance model does fit reasonably well on the MultiNav Pro+ once you've got your 1m RSSI baseline nailed in-situ like Thomas said - I'm getting usable results with n around 2.7 in my indoor environment. The median filter tip from Sarah is solid, that alone cleaned up my estimates more than anything else I tried.